Professional Documents
Culture Documents
The Optimization of Rifle Barrel Harmonics
The Optimization of Rifle Barrel Harmonics
The Optimization of Rifle Barrel Harmonics
UVM ScholarWorks
2022
Recommended Citation
O'Neil, Daniel, "The Optimization of Rifle Barrel Harmonics" (2022). Graduate College Dissertations and
Theses. 1608.
https://scholarworks.uvm.edu/graddis/1608
This Thesis is brought to you for free and open access by the Dissertations and Theses at UVM ScholarWorks. It
has been accepted for inclusion in Graduate College Dissertations and Theses by an authorized administrator of
UVM ScholarWorks. For more information, please contact schwrks@uvm.edu.
The Optimization of Rifle Barrel Harmonics
A Thesis Presented
by
Daniel O’Neil
to
of
August, 2022
From the optimization of mechanical systems perspective, firearms offer a plethora of complex
scenarios. Many variables, coupled with sometimes unstable systems, must be balanced appro-
priately by the user to achieve precision, that is to say, to hit the target bullseye from some
distance away with a small projectile launched on a parabolic trajectory. The sources of error
entirely affect final terminal precision, some of them related to human factors, others related to
the quality of the ammunition, and the remaining associated with the mechanical capability of
the weapon. This thesis documents the vibrations analysis of an isolated rifle barrel, identifies
barrel tip displacement and angle of rotation in firing scenarios both analytically and empirically,
and makes recommendations on optimizations and control schemes to improve overall precision
capability.
Table of Contents
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
5 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
6 Barrel Optimizations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
8 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
ii
List of Figures
iii
27 ANSYS Modal Analysis of Tikka T3X System, Mode 1, 38.6 Hz . . . . . . . . . . 29
34 Tikka System Rifle Barrel Test, Time Series Acceleration Data, Test 1, Shot 1 . . 35
35 Tikka System Rifle Barrel Test, Empirically Determined Modes, Test 1, Shot 1 . 36
36 Tikka System Rifle Barrel Test, Time Series Acceleration Data, Test 1, Shot 2 . . 36
37 Tikka System Rifle Barrel Test, Empirically Determined Modes, Test 1, Shot 2 . 37
39 Tikka System Rifle Barrel Test, Time Series Acceleration Data, Test 2, Shot 1 . . 39
40 Tikka System Rifle Barrel Test, Time Series Acceleration Data, Test 2, Shot 2 . . 40
42 Short Barrel Optimization vs. Tikka T3X System Action Time Response . . . . 44
43 12.5” Mid Gas System in 5.56mm (L) and 12.5” Piston System in 5.56mm (R) . 45
44 Grid Style Extruded Barrel Blank Model, Area Moment of Inertia Optimization 46
47 Moment of Inertia Optimization vs. Tikka T3X System Action Time Response . 49
iv
List of Tables
v
1 INTRODUCTION 1
1 Introduction
The use of tools to more efficiently perform a task, with few exceptions, is a decidedly human trait.
Throughout history, humans have continued to invent, improve, design, iterate, and otherwise
engineer tools to achieve goals and broaden overall technological capability. Of particular interest
is in optimizing the modern rifle barrel for the purposes of improving precision and thereby
a reader who is skilled in the art may understand, a mechanical assembly such as a rifle is
susceptible to the laws of dynamics just like any other mechanism, and there are inefficiencies
or minimizing these sources of error that it is possible to achieve greater precision in a mechanical
system such as a rifle. This thesis explores the mathematical options for optimizing rifle barrel
performance through minimizing error. One source of error is the way the barrel behaves during
a firing event. Consider the rifle(1) shown in Figure 1 in caliber 6.5 Creedmoor, which is a
6.5mm diameter projectile traveling at 2650 feet per second and with a weight of 130 grains or
9.07 grams.
1 INTRODUCTION 2
This is an example of a modern bolt action rifle, and is representative of most commercially
available sporting rifles. The rifle barrel itself is ”free floated”, meaning that it does not contact
the firearm stock at any point along its length from the action to the muzzle. This degree of
separation ensures the barrel can vibrate freely and naturally, without influence by external forces
through stock contact, and perform repeatably for each firing event.
When a cartridge is loaded into the rifle and fired through the barrel as designed, the firing pin
hits the cartridge primer, which ignites the smokeless powder (propellant) internally contained in
the cartridge. The rapid burning of the propellant increases pressure severely until the projectile
is forced out of the cartridge and down the barrel at a predetermined velocity based on projectile
(2)
As the projectile is forced out of the cartridge and into the rifling (Figure 3), the rifling
engraves the projectile body and imparts spin. Barrels are typically designed with a consistent
twist rate e.g. 1 rotation in 8 inches of travel or 1:8. With larger cannon barrels, it is common
to have a variable twist rate or ”gain twist”, where the projectile spin rate increases the further
it travels down the barrel and reaching a max spin at the muzzle end. This paper will consider
only barrels with consistent twist rates, although the same principle could be applied to gain
twist barrels.
(3)
1 INTRODUCTION 4
Optimal barrel twist rate for a given projectile may be calculated using the Green Hill For-
mula, given bullet diameter, length, specific gravity, and constant C (C=150 for projectile veloc-
r
C ∗ D2 Sg
T = ∗ (1)
L 10.9
r
150 ∗ .2642 10.05 inches
T = ∗ = 7.27 (2)
1.380 10.9 rotation
For this projectile, a 1:7.27 twist rate would be optimal, however, commercially, barrels are
produced in a 1:8 twist rate to handle a larger variety of available projectiles weights. While
the projectile is traveling down the barrel, it is influenced by the burn rate of the propellant,
and the subsequent pressure developed during the firing event. This it typically captured in
a pressure curve, which peaks while the projectile is still traveling down the barrel bore. A
representative pressure cure is shown below in Figure 4(4). It should be noted that the projectile
is only influenced by barrel movement while it is inside the rifle barrel. After it exits, the barrel
may move and vibrate extensively however, it is irrelevant from the perspective of the projectile
For the short time that the projectile travels within the barrel bore, it is subjecting the barrel
to reactionary loads, and follows the movement of the barrel as it vibrates at its natural modes.
Loads reacted by the barrel, especially for imperfect projectiles or barrels, and natural vibration
will cause deviations between the point of aim, and the point of projectile impact.
2 Literature Review
An extensive amount of work has already been accomplished on rifled barrels for both commercial
and military purposes. Mr. J. Siewert points out that ”Barrel-pointing variability accounts for
85%-95% of the barrel-related dispersion error budget...”(21). As far back as the early 1900s,
engineers and technicians were working to understand the dynamic movement of a rifle barrel
and indeed, of the weapon itself, for the purposes of improving precision and system design.
On June 6, 1901, Mr. A. Mallock by way of Lord Rayleigh, F.R.S., communicated a paper
that discussed the phenomenon of rifle ”flip” attributed to the method of which a rifle barrel is
attached to its stock. That is to say, the rifle barrel center axis vertical offset from the point
2 LITERATURE REVIEW 6
at which the stock contacts the user’s shoulder, causing the muzzle to offset from the aimpoint
during a firing (recoil) event. Mr. Mallock proposed rifle mode shapes based on the firing
excitation, derived rudimentary equations of motion and that which described response, and
measured natural frequencies based on a Lee Enfield Rifle(12). Mallock concludes that muzzle
tip slope at the moment of projectile exit and cartridge velocity variation as having a distinct
effect on weapon precision. Mallock expresses in his paper that a ”good modern sporting rifle
will shoot within a probable deviation of considerably less than 2 feet from the intended path.”
Modern weapons and ammunition have made great strides since then!
More modern studies tend to be of the form of a Finite Element Analysis (FEA), checked
against classical undamped equations of motion. Xu, Guan, Liu, and Xu in their paper ”Study
on Barrel Vibration Characteristics of Typical Sniper Rifle”, use a traditional FEA approach to
2 LITERATURE REVIEW 7
analyze a sniper rifle barrel, develop modeshapes and displacement, and then compare results to
empirically collected data using a high speed camera. Their model to empirical data correlation
This was similarly accomplished by Pyka, Bocian, Jamroziak, Kosobudzki, and Kulisiewicz in
the paper ”Concept of a Gun Barrel Based on the Layer Composite Reinforced with Continuous
Filament”. In this case, the team developed an FEA model of a subject machine gun barrel,
collected test data on that barrel, and then proposed a carbon fiber exterior sheath with which
to augment barrel stiffness and therefore shift natural frequency in a mechanically favorable
direction (14).
Figure 6: Carbon Fiber Wrapped Machine Gun Barrels, Pyka et al., 2019
work on ”Analysis and Control of Gun Barrel Vibrations, he delves into the factors affecting
tank barrel motion and types of vibrations sustained in an armored vehicle. His data includes
accelerometer data collected from live tank operations and from a fixtured shaker armature to
support his hypothesis. Model development consistently utilizes FEA to approximate a real
world system, and develop mode shapes and displacements. His solution is a ”Tuned Vibra-
tion Absorber”, or oscillating single degree of freedom system that cancels or reduces vibration
2 LITERATURE REVIEW 8
He also published a second work titled ”A Review on the Gun Barrel Vibrations and Control
for a Main Battle Tank” in cooperation with Dursan and Utlu, where they explored the relation-
ship of tank cannon muzzle length to its anticipated displacement. He finds a direct relationship
between barrel length and accuracy of that combat vehicle, especially for vehicles that are not
static during a firing event. Their solution was a Dynamic Muzzle Reference System (DMRS)
that controlled aiming offset and stabilized muzzle vibrations during the event to maximize pre-
cision of the cannon barrel. They also recommend shorter gun barrels (relative stiffness) and
Figure 7: Tank Armored Fighting Vehicle Dynamic MRS, Dursun et al., 2017
Similarly, Koc, Esen, and Cay from Sakarya University published a paper on large caliber
cannon barrels utilized in armored vehicles, specifically the effect of unbalanced projectiles and
their related error. In their paper, they again define equations of motion particular to a tank
cannon for each axis of oscillation, generate modal mass, stiffness matrix, and damping relative
to barrel elements, and then generate cannon barrel predicted response. The purpose of their
effort seems to be the utilization and validation of a novel analysis tool named 12 DOF FEM
Mr. G. Kolbe developed an analysis and method for tuning rifle barrels to an optimum
point for precision shooting at a known distance in his paper ”Using Barrel Vibrations to Tune
2 LITERATURE REVIEW 9
a Barrel”. After a review of prior work both reputed and infamous (to include referencing
Mr. Mallock’s work from 1901 above), he discusses measurement of barrel motion in one axis
(vertical), and the measurement of the exact moment the projectile leaves the muzzle by using
what he terms a ”muzzle gate”. In this way, Kolbe collects muzzle angle, and muzzle velocity,
which previously had only been derived using position, velocity, and acceleration mathematical
models. He concludes with demonstrations of firing events with an untuned barrel firing groups
at 50 yards, and a barrel tuned with a mass at the end to control projectile exit timing. With
consistent projectile exit, Kolbe is able to significantly increase rifle precision. It is apparent that
this tuning must be accomplished any time ammunition load or projectile weight change (18).
The company TacomHQ has developed a product termed ”Structured Barrels” that they
market as harmonically dead (19). The actual product is a large diameter barrel blank that is
gun drilled and rifled along the center axis, as a normal rifle barrel would be, but it is then gun
drilled concentrically to form tubes around the barrel. TacomHQ offers some explanation and
analysis to support their claim. It appears that they have optimized a geometric shape, as it
relates to its cross sectional moment of inertia, for increased stiffness, lower weight, and a natural
There is also a product sold under the company ”Recreational Software, Inc”, that allows a
user to measure exact moment of exit of the projectile from a rifle barrel. This software guides the
user in developing customized reloaded cartridges that take advantage of a specific rifle’s barrel
behavior during a firing impulse. It refers to muzzle ”whip” and barrel ”ringing” to describe
In Mr. Potts’ ”Things that affect rifle accuracy (and how to correct them)” article for Shootin-
gUK, he discusses rifle vibrations as a significant source of error. ”The cause of more inaccuracies
and barrel problems than realized. The barrel vibrates as the bullet passes along its length, so
it is best to keep that vibration as smooth and uniform as possible.” He advocates for different
types of rifle action bedding, that is, a way to stabilize and support the rifle action with sub-
stances like epoxy. He also mentions the importance of having a free floated barrel that is not
Referencing Calucci and Jacobsen(5), a perfectly homogenous and symmetric barrel will expe-
rience a predictable array of loads as the projectile travels down the bore. ”During and after
firing, the unbalanced forces on the gun may be categorized as the gas force FR , the projectile
resistance force FP r , and the rifling force FT ”(5). Figure 10 below shows where these loads occur.
The rifling force may be related to the projectile force via the following equation:
k
FT = ( d )2 ∗ FP r ∗ tan(α) (3)
2
Where k is the projectile radius of gyration, d is projectile diameter, and α is the angle of the
rifling. Notionally, for a perfect rifle barrel firing a perfect projectile, there is no force unbalance
reacted by the barrel, and thus no real impact to overall system precision at the moment the
projectile exits the muzzle of the barrel and thereby out of its influence. With an asymmetric
rifled bore, it is apparent that a force component of some magnitude would act on the rifle barrel
in such a way as to effectively generate a moving load traveling at the point of projectile rifling
contact perpendicular to the barrel wall. This force would be the rifling force FT at its peak due
Let us shelve this concept for a moment and explore the mathematics that govern barrel
movement during a firing event. For the purposes of developing a model, the barrel is assumed
3 BARREL LOADS AND NATURAL VIBRATIONS 12
to behave elastically such that the equations of motion for a beam may be used. In this case, the
boundary conditions are for a fixed-free configuration, and will be stated plainly below.
Firstly, let us start with the general equation of motion for a classically damped system:
Equation 4 above is for a single degree of freedom system, subjected to a forcing function on
the right side of the equation. Although only for a single degree of freedom, this equation may
be leveraged to analyze n degrees of freedom (DOFs) through the process of Modal Analysis.
In this case a more appropriate rendition of the equation for n degrees of freedom is the matrix
format:
Where the [M ] is the DOF lumped mass matrix, [C] is the damping matrix, and [K] is the
stiffness matrix for the system. System free response and eigenvalues are determined from the
Considering our example system, the Tikka T3X 6.5 Creedmoor rifle, we can develop a math-
ematical model for barrel position relative to an axis coincident and parallel with the bore
barrel length. In this way, each nodal displacment motion may be quantified.
It is important to select an adequate number of degrees of freedom for this method of calcu-
lation, as too few will result in noticeable error from the discretization. The first iteration for
this paper only contained 6 DOF and exhibited approximately 15-20% error when compared to
Finite Element Analysis results. Let us consider 10 nodes and thus 20 degrees of freedom, un-
derstanding that each node will exhibit displacement and rotation. Utilizing beam displacement
and rotation shapes from Tedesco (6), it is possible to develop the stiffness matrix one degree of
freedom at a time. Degree of Freedom 1 and reaction moments and loads are displayed below in
3 BARREL LOADS AND NATURAL VIBRATIONS 13
Figure 11.
This is possible by applying a displacement of magnitude ”1” at the node of interest, and
fixing displacement of all other nodes in the system to 0. At this point, simple beam equations
apply at node 1. The end result is the stiffness vector {K1 }, which is a column vector of length
24EI −12EI 6EI
K1 = , , 0, 0, 0, 0, 0, 0, 0, 0, 0, 2 , 0, 0, 0, 0, 0, 0, 0, 0 (6)
L3 L3 L
Likewise for degree of freedom 2 at Node 2, the corresponding beam shape and stiffness vector
{K2 } is:
−12EI 24EI −12EI −6EI 6EI
K2 = , , , 0, 0, 0, 0, 0, 0, 0, , 0, 2 , 0, 0, 0, 0, 0, 0, 0 (7)
L3 L3 L3 L2 L
This pattern continues through degree of freedom 10, which completes the displacement
3 BARREL LOADS AND NATURAL VIBRATIONS 14
lbs
shapes of each node. For the Tikka T3X barrel, utilizing Esteel = 30 ∗ 106 in 2 , OD = .75in,
π 4
ID = .22in, I = 64 (OD − ID4 )in4 and L = 23.5/n inches, the first 10 DOFs for the stiffness
matrix are shown below. Note that the first 10 DOFs are displacement only, and the final stiffness
It should also be noted that the stiffness matrix is always symmetric. After applying beam
formulas for each of the displacement degrees of freedom, the same must then be repeated for
rotations at each node for 11-20th degrees of freedom. Similar to the displacements applied above
to generate stiffness values, now rotations are applied at each node and reaction moments and
forces are calculated. All other nodes have rotation fixed at 0, and the node of interest has a
rotation (moment) applied. For node 11, the subsequent beam rotation shape(6) and reactions
are:
−6EI 8EI 2EI
K11 = 0, , 0, 0, 0, 0, 0, 0, 0, 0, , , 0, 0, 0, 0, 0, 0, 0, 0 (8)
L2 L L
This process continues on to node 12 in the same fashion, generating the beam rotation shape
as follows:
6EI 6EI 2EI 8EI 2EI
K12 = , 0, , 0, 0, 0, 0, 0, 0, 0, , , , 0, 0, 0, 0, 0, 0, 0 (9)
L2 L2 L L L
Progressing through each displacement and rotational degree of freedom allows the develop-
ment of the system stiffness matrix, dependent only on E, I, and beam length L. Beam length
is the linear distance between nodes on the beam, and is the resolution of the discretization that
The resulting overall stiffness matrix with all degrees of freedom for 10 nodes is:
3 BARREL LOADS AND NATURAL VIBRATIONS 16
Figure 16: 20 Degree of Freedom Stiffness Matrix K, for Tikka T3X Rifle
Going back to the classical equation of motion below, we can now plug in the stiffness matrix,
leaving the damping matrix [C] and the mass matrix [M ] still undefined. The mass matrix is
simple in that it is a diagonal matrix with each mii value equal to the overall mass of the barrel
divided over the number of discretized segments. The damping matrix [C] is normally estimated
during mathematical modeling, or determined empirically through test, which will be covered
later in this thesis. For the purposes of this mathematical model, the damping value was assumed
Because rotational degrees of freedom do not have a mass associated with them, they may be
disregarded by conducting a Static Condensation of the stiffness matrix K. It follows that the
stiffness matrix [K] may be dissected into its displacement and rotational components per(6):
Ktt Ktθ X Ft Ft
= =
Kθt Kθθ θ Fθ 0
Ktt Ktθ
=K
Kθt Kθθ
3 BARREL LOADS AND NATURAL VIBRATIONS 17
degree of freedom. The force subvectors Ft and Fθ correspond to the displacement subvectors X
and θ. If force subvectors in the system do not include rotational components, then Fθ = 0, and
the stiffness matrix may be condensed. In this case, evaluating the rotational displacements per
−1
θ = −Kθθ Kθt X (11)
−1
(Ktt − Ktθ Kθθ Kθt )X = Ft (13)
Therefore, the overall stiffness matrix equation shown above may be represented more simply
Kt X = Ft (14)
−1
Kt = Ktt − Ktθ Kθθ Kθt (15)
In this way, only the degrees of freedom with associated mass are represented in the condensed
matrix (10x10) which also correlates to the sizes of both [M ] and [K]. This extensive matrix
With [K] and [M ], we are now interested in the modeshapes and natural frequencies of the
system. This is achieved by solving the Eigenproblem for both Eigenvectors and Eigenvalues.
Knowing the equations of motion for the system (Equation 4), and disregarding any system
damping, we can set it equal to zero and postmultiply by the inverse of the mass matrix ([M ]−1 )
to yield (6):
Since:
3 BARREL LOADS AND NATURAL VIBRATIONS 19
then:
Which is recognized as the standard eigenvalue problem. Its solution results in an eigenvalue
diagonal matrix of natural frequencies for each degree of freedom, and an eigenvector matrix of
modeshapes. The solution to which is simplified by utilizing the eig function in MATLAB, and
Although the stiffness matrix and corresponding Eigenvalue problem was developed indepen-
dent of the forcing function F (t), we now must consider how damping and the forcing function
impact the system’s vibration response during a firing event. If the forcing function remained
static at one point of the barrel, the solution and subsequent system response would be trivial.
However, the projectile in this case is not stationary. The forcing function, for the purpose of
this mathematical model, was set up as an asymmetric load on the rifling, traveling down the
barrel with the projectile. In order to correctly build this model, Warburton was consulted on
moving loads in vibrational systems(7), and Blevins was used for modeshape equations(8).
At t = 0, it is assumed that the projectile is still unfired and in the chamber, and no motion
has yet started. At t corresponding to the travel of the projectile down the barrel distance x at
is:
L 23.5in
ti = = in
= .00074sec (20)
U 31800 sec
However, this erroneously assumes an instantaneous velocity, and does not account for the
realistic time for primer strike, combustion and pressure to build, and projectile exit from the
3 BARREL LOADS AND NATURAL VIBRATIONS 21
muzzle. This term is called action time, and will be assumed to be 0.002 seconds, or a corre-
in
sponding average velocity of 11, 750 sec , in line with similar cartridge industry empirical data.
Warburton sets up the moving forcing function (constant force) for a simply supported beam
as(7):
1 sπU T
φs (a) = 2 2 ∗ sin (21)
l
Where s is the mode (s=1, 2, 3,...), U is projectile velocity, and t is time of projectile travel
1
(starting from 0). The 2 2 is a normalizing factor particular to his solution for a moving force on
1 P sin(sπ Ul t )
qs = Bs sin(ωs t) + Cs cos(ωs t) + 2 2 (22)
m ωs2 − (sπ Ul t )2
If the beam starts at rest when t=0, then initial conditions and boundary conditions drive
1 Ut U
2 2 P sin(sπ l ) − (sπ lωs )sin(ωs t)
qs = (23)
m ωs2 − (sπ Ul )2
Since gunfire provides broadband excitation, we assume that the forcing function occurs at
the natural frequency of the mode being excited in the Tikka T3X system. Equation 23 may
1
22 P
q¨s + ωs2 qs = sin(ωs t) (24)
m
P
qs = 1 (sin(ωs t) − ωs tcos(ωs t)) (25)
2 mωs2
2
However, this is inadequate to describe the function for a cantilever beam such as a rifle
barrel; instead the case needed is fixed-free. Referring to Blevins modeshapes, it is understood
3 BARREL LOADS AND NATURAL VIBRATIONS 22
that a fixed-free system’s (Blevins labels it a ”clamped-free” system) modeshapes may be math-
λi x λi x λi x λi x
ρi = cosh( ) + cos( ) − σi [sinh( ) + sin( )] (26)
L L L L
Where λi specifically for Equation 26, is the following vector, corresponding to mode(8):
beam, we also know the modeshapes are sinusoidal. For a fixed-free system, the sin in Equation
24 was replaced with the modeshape equation in Equation 26, and then utilized to derive the
1
2 2 P ρi
M Fi = (29)
µii
In the modal force equation above (Equation 29), P is the magnitude of the moving force, ρi is
the modal influence (Blevins modeshape), and µii is the modal mass value for the ith node. The
modal mass matrix [µ] is necessary to achieve proportional values in the response calculation,
T
[µ] = {ψ} [M ] {ψ} (30)
At this point, the moving force for each mode, M Fi , may be calculated, which results in the
right side of the modal equation shown above. Set up for the Tikka T3X fixed-free system, the
1
2 2 P ρi
q¨s + ωs2 qs = = M Fi (31)
µii
This modal force is applicable only while the projectile is traveling down the barrel. At all
other points in time, the modal force on the rigth side of the equation is 0. We know that this
modal equation of motion was derived using a modal transformation of the form:
[Ψ]T [M ][Ψ] {q̈} + [Ψ]T [C][Ψ] {q̇} + [Ψ]T [K][Ψ] {q} = M Fi (32)
The next question is how to solve Equation 31? A direct integration is possible, but cumber-
some and nontrivial. It is more appropriate to solve using the convolution integral or Duhamel’s
Due to the principle of superposition, meaning that each segment of the the response inte-
gration may be added together to obtain the complete response, the Duhamel integral may be
Z t
x(t) = F (τ )h(t − τ )dτ (34)
0
In order to solve Equation 31, we must utilize Duhamel’s integral to convolve the impulse
The impulse response h(t) is easily calculated from the characteristics of the dynamic system
1 −ζωt
h(t) = e sin(ωd t) (35)
mωd
Because the principle of superposition was used to establish Duhamel’s integral, we may apply
3 BARREL LOADS AND NATURAL VIBRATIONS 25
it to this linear system and determine the response due to excitation from a gunfire event(6).
Z t
1
x(t) = F (τ )e−ζω(t−τ ) sinωd (t − τ )dτ (36)
mωd 0
The result of evaluating this convolution integral, and multiplying by the modal force, is in
modal coordinates and must be transferred back to physical coordinates by multiplying by the
modeshape matrix ψ which gives the equation of motion q for each mode. Again due to the
principle of superposition, the total response may be obtained by adding up the q value for each
This process was accomplished for modes one and two in this mathematical model, and the
The first three modes of the Tikka system occur at 35.7, 224.8, and 632.2 Hz respectively
A double check utilizing the Blevins calculations for fixed-free beam natural frequency result
in:
λi EI 1
fi = ( )2 (37)
2πL m
Where λi = [1.87510407, 4.69409113, 7.85475744, ...] according to the values specified in Blevins
Modeshapes(8).
The Blevins calculation results in the first three modes occurring at 39.3, 246.0, and 688.9
ANSYS was also utilized to correlate the analytical models to the subsequent empirical data. For
the ANSYS model, the rifle barrel was represented exactly to dimension, as a 23.5” long tube
with inner and outer diameter’s to match the actual rifle barrel geometry. A pre-stressed Modal
Analysis method was used to simulate a force moving down the barrel. The receiver end of the
barrel was fixed by constraint. The barrel itself was constrained with a remote displacement
to allow only displacement in the Y axis (vertical component), and rotation of each degree of
freedom about the X axis. Z axis deflection and oscillation were considered negligible for this
analysis.
Using the Static Structural Module to create the pre-stressed environment, a force was applied
to each discretized segment of the rifle barrel in steps, with the force moving to the next segment
with each step increment. The Static Structural analysis was then fed into a Modal Analysis
Module for the final modal calculations of modeshape and natural frequencies.
With constraints applied, the moving force function was arranged to shift to the next beam
The Modal Analysis was set up to calculate modal displacement. The magnitude of deflection
is not true to the actual system; it is scalable by any factor, similar to the modeshapes matrix
Figure 27: ANSYS Modal Analysis of Tikka T3X System, Mode 1, 38.6 Hz
4 ANALYTICAL MODELING RESULTS USING ANSYS 30
Figure 28: ANSYS Modal Analysis of Tikka T3X System, Mode 2, 240.7 Hz
Figure 29: ANSYS Modal Analysis of Tikka T3X System, Mode 3, 669.8 Hz
The ANSYS Modal Analysis for the first three modes of the Tikka T3X system resulted in nat-
ural frequencies of 38.6, 240.7, and 669.8 Hz. Modeshapes matched the discretized mathematical
model within 7%, and also the Blevins calculations(8) to within 3%.
5 EMPIRICAL RESULTS 31
5 Empirical Results
Before proposing any changes to the Tikka system, especially as an improvement, it is important
to correlate the models developed with real world empirical data. To that end, a Tikka T3X rifle
was utilized to collect live fire data with a DEWESoft Sirius (10) Data Acquisition (DAQ) System
and PCB Accelerometers. It proved to be difficult to collect valid data due to high acceleration
loading during firing events, and resultant low data resolution which prompted a second live fire
event. The data and conclusions presented below are arranged in the order easiest for the reader
to follow, and not in the order in which the data were collected.
Testing was set up at a private range facility, and utilized twin +/-500G accelerometers
mounted to the rifle barrel. The action time for primer strike to projectile exit was assessed as
occurring within 2 milliseconds, which drove the choice for channel sample rate. This equated
1 1
to an event frequency of actiontime = .002 = 500Hz. In terms of signal processing, the Nyquist
Theorem dictates that, in order to capture frequency content with known error in peak magnitude,
we must sample at least twice the frequency of event occurrence. Sampling faster than 2x
increases the accuracy of the peak magnitudes, for example, at the dominant frequencies. The
more oversampled the data set, the more accurate the frequency content magnitudes, and the
smaller each frequency bin may be (x axis resolution) for a frequency domain transform. Common
Figure 30: Tikka System Rifle Barrel Test, DEWESoft DAQ, PCB Accels
In order to understand free vibration modes and magnitudes, a tap test was performed on the
rifle barrel. Accelerometers were placed at the muzzle and at 5 inches from the muzzle towards
the breech end of the barrel. The accelerometers were secured to the barrel using hose clamps;
one placed at the muzzle end, and one placed 5 inches back towards the breech end from the
muzzle. A sample rate of 100kHz was selected for the tap test to ensure data resolution during
post processing. The DAQ system was set up to apply an antialiasing Butterworth filter at
50kHz. Higher order filters are normally used when sharp data attenuation is desirable. Figure
(20)
With the system live and collecting data, the barrel was rapped sharply with a steel screw-
Figure 32: Tikka System Rifle Barrel Test, Tap Test, 100kHz
5 EMPIRICAL RESULTS 34
The tap test results were analyzed within a 0.002 second window of time, and cleanly show the
impulse, free vibration of the system, and the modes of oscillation in the Fast Fourier Transform
(FFT). Data was plotted up to 500 Hz as the window of interest since high frequency modes
were orders of magnitude lower in amplitude. Note that the first and second barrel modes are
dominant in the FFT, along with other modes excited within the system. This is expected since
a complete rifle system is the subject of this test, not just the rifle barrel.
The first live fire test mimicked the set up for the tap test, accelerometers were placed at the
muzzle end and at approximately 3 inches aft from the muzzle end towards the breech end. Ac-
celerometer rigid wax was used to fix the accelerometers in place, and held consistent throughout
each round fired, although lower magnitude accelerations resulted in this measurement iteration.
This is assessed as due to the coupling between the barrel and accelerometers; the accelerometer
wax was not as stiff as the hose clamps used for the tap test and second live fire test. The rifle
itself was held rigidly by a Pig Saddle and aluminum field shooting tripod from Shadow Tech(9).
The rifle was man fired for each data set collected. Again, the DEWESoft DAQ system was set
up for two channel data collection, this time at 20kHz per channel, with an antialiasing filter
Figure 34: Tikka System Rifle Barrel Test, Time Series Acceleration Data, Test 1, Shot 1
Accelerometer data is windowed from the moment of sear release (the moment the trigger
is pressed) to 0.002 seconds later, this being the action time of the system. Data resolution is
immediately apparent since only approximately 38 data points are available for review. Acceler-
ation magnitudes are within the accelerometer capability. The accelerometer channels were then
integrated twice to obtain displacement. The actual integration algorithm was managed internal
to the DEWESoft platform, and was high pass filtered below 0.1Hz to reduce drift and error.
Total barrel displacement for the action time window is 0.007 for the muzzle and 0.002 for the
The time series acceleration data was then transformed to the frequency domain by means of
an FFT within DEWESoft. Filtering was implemented according to DEWESoft best practices
and the author’s prior data collection knowledge. That is to say, a band pass filter was imple-
mented from .1 Hz to 10 kHz to attenuate the transform DC offset at low frequency and high
frequency noise.
Figure 35 below displays the empirically determined modes of oscillation for the Tikka T3X
Figure 35: Tikka System Rifle Barrel Test, Empirically Determined Modes, Test 1, Shot 1
Shot 2 was performed in exactly the same manner, and the subsequent data is shown below
in Figures 36 and 37. Again note that the time window was 0.002 seconds from sear release.
Figure 36: Tikka System Rifle Barrel Test, Time Series Acceleration Data, Test 1, Shot 2
5 EMPIRICAL RESULTS 37
Figure 37: Tikka System Rifle Barrel Test, Empirically Determined Modes, Test 1, Shot 2
The modes from shot 2 are consistent with that from shot 1, with both the first and second
After reviewing the live fire data from test one, the determination was made that the data were
insufficient to proceed, primarily based on the resolution of the frequency content. The assessment
was that the frequency data did not have enough frequency bins for the x axis to adequately
determine resonant modes and peaks. It was also desirable to have higher resolution data to
reduce error for the double integration to obtain displacement. To that end, the DEWESoft
DAQ system was again secured and set up again for a two channel system (vertical barrel motion
only), sampled at 100 kHz per channel. In this way, the data were amply oversampled to assure
decent frequency content and magnitude for later use in system optimization.
Setup and accelerometer filters were adjusted slightly to optimize the Butterworth bandpass
filter applied for anti-aliasing from .8 Hz to 30 kHz. The filter was implemented using a 6th
order algorithm to accentuate the filter rolloff and data attenuation closer to the filter edges.
The accelerometers were fixed to the barrel using hose clamps for greater rigidity and coupling.
Accel 1 was located at the muzzle end, and Accel 2 was at -5 inches from the muzzle back towards
Two shots were expended for test 2, and even with +/- 500G accelerometers, the sensors
experienced overload errors and railed at measured magnitudes (the sensor signal input went
beyond the capability of the sensor), and the resultant frequency content was erroneous. Upon
further investigation, the accelerometers measured accurately during the action time of the sys-
tem, but grossly exceeded their capability once the projectile left the barrel and the totality
of the impulse response played out. A specific example of this is displayed below in Figure 38.
Note the accelerometer time series magnitude offset with corresponding large y axis value outside
the calibrated range of the device. The frequency versus acceleration plot shows low frequency
content with a progressive attenuation down to zero magnitude, in no way corresponding to the
impulse that the system experienced during the firing event. While interesting, this data is not
usable for this analysis except to help select higher range accelerometers for future testing.
The author asserts that the acceleration data within the action time of the system is still
valid, but the follow on accel data is garbage and cannot be used. The double integration of
the signals in shot 2 used for displacement resulted in values that increased to a nonsensical
order of magnitude before dropping off again; clearly laden with error related to saturating the
time series was error fraught, with extensive drift and offset. See Figures 39 and 40; the asterisk
Figure 39: Tikka System Rifle Barrel Test, Time Series Acceleration Data, Test 2, Shot 1
5 EMPIRICAL RESULTS 40
Figure 40: Tikka System Rifle Barrel Test, Time Series Acceleration Data, Test 2, Shot 2
2 2 .030* .030* 0*
Muzzle angle is derived from the measured slope between the max displacement of each
accelerometer. In other words, it is the slope between two points of the modeshape at the end
of the barrel. Note that the error at 100 yards, or 3,600 inches of distance equates to a vertical
delta of 7.67 inches for a muzzle angle of 0.122 degrees, and half that for the muzzle angle of 0.10
degrees. Obviously, this error compounds at distance. Normal midwest shots for hunting big
game exceed 600 yards distance, which equates to an 46 inch error at the target, solely due to
muzzle angle and displacement. In military engagements, marksmen may engage targets beyond
1,500 yards, which in this case equates to a 115 inch error at the target independent of nominal
6 BARREL OPTIMIZATIONS 41
projectile drop and wind calls. If an average human is 20 inches across at the shoulders and 6
feet tall, this error is obviously significant and will result in a target miss without correction.
From the data collected, several items are apparent. The first is that there are modes from
the entire system oscillating besides just the barrel modes. As mentioned prior, this is expected
since the complete rifle is fitted with an optic and is shooting from a rigid tripod mount. It is
however, obvious that the first two barrel modes predominantly impact system oscillations, and
in some of the empirical test cases, dominate the vibration energy content. Second is that the
empirical displacement data aligns with the first barrel modeshape at 36 Hz for all correlated
shots. For shots where the accelerometers exceeded their magnitude capability or exhibited
response errors, that data was discounted. Last is that the tap test data validates the frequencies
and modeshapes of the analytical (FEA) and mathematical models to within 10%. Moving
forward, the mathematical and analytical models will be used to suggest optimizations for the
rifle system, predominately targeting the first and second barrel modes.
6 Barrel Optimizations
There are several opportunities for altering and/or optimizing barrel vibrations. They will be
presented here in order of complexity. In this case, barrel optimization equates to better precision,
or less error at the target distance, which is directly tied to muzzle tip displacement and angle
of rotation at projectile exit. Both ANSYS and the MATLAB mathematical model are used to
demonstrate the various aspects of barrel improvements. In the case of the MATLAB model, the
data is normalized to the Test 1 data magnitudes. Using this algorithm, the stock Tikka T3X
An easy and obvious way to improve barrel deflection during firing events is to make the barrel
shorter. From the beam shape equations (6), we know that barrel length exponentially affects
12EI
F = (38)
L3
6EI
F = (39)
L2
By lowering the overall barrel length, and thus exponentially reducing the denominator of the
reaction forces equation, we achieve a stiffer barrel as defined by the new stiffness matrix [K].
In the example of the Tikka T3X system, the mathematical model predicts first and second
modes at 35.7 and 224.8 Hz for the 23.5 inch length barrel. Mode 1 is predicted to cause .0016
inches of deflection at the muzzle, and mode two will cause .0055 inches of deflection at the
muzzle. The predicted angular error at the muzzle tip from the second mode is .077 degrees,
which equates to a vertical error of 4.83 inches at 100 yards or 48.3 inches at 1000 yards.
Now, if the barrel length is adjusted to one half of that length, we see an immediate improve-
ment in the system’s resonant modes. Mode 1 is shifted from 35.7 Hz to 143 Hz, and Mode 2 is
shifted from 224.8 Hz to 899 Hz. Muzzle displacement drops by approximately 60% from .0055
inches to .0021 inches when excited at Mode 2. And it follows that the shorter, stiffer barrel,
likewise has a lower muzzle tip angle at the cusp of projectile exit resulting in lower targeting
error at distance.
Figure 41: Short Barrel Optimization vs. Tikka T3X System Response
6 BARREL OPTIMIZATIONS 44
Figure 42: Short Barrel Optimization vs. Tikka T3X System Action Time Response
This seems like an easy fix; why not make all rifle barrels shorter? The unfortunate trade
off is that interior ballistics demands a certain barrel length to achieve powder burn efficiency,
pressure curves, length of barrel to stabilize the projectile, and a muzzle velocity that is not
severely impacted as to prevent long distance shooting in the first place. From the end user’s
perspective, shorter barrels are not always bad, but lower muzzle velocity and thus terminal
energy at the target range, equate to less lethality, projectile stability (supersonic, transonic, and
subsonic), and overall accuracy. Barrel lengths must be chosen carefully to optimize not only
barrel harmonics, but also muzzle velocity, projectile performance, and overall system weight.
Figure 43 shows two industry prototype options that the author has successfully shot out to 1100
yards with standard ammunition, demonstrating that long barrels are not always necessary to
6 BARREL OPTIMIZATIONS 45
shoot precisely.
Figure 43: 12.5” Mid Gas System in 5.56mm (L) and 12.5” Piston System in 5.56mm (R)
Another method to optimizing barrel performance is to increase its Area Moment of Inertia, or
variable I used in the beam stiffness equations. Adjustments to I impact the overall stiffness
matrix [K] as part of the equations of motion, and subsequent reaction loads for each degree of
freedom; the higher the Area Moment of Inertia, the greater the beam stiffness, the lower the
Manufacturers make barrel profiles currently that are more complex than simply round or
tapered cylinders. Profiles include hexagonal, diamond, and fluted. Fluting is marketed as an
improvement for both barrel stiffness and heat transfer, although improvement of these claims is
dubious at best. There is also the example cited earlier in this paper regarding the structured rifle
barrels and their claim of being ”harmonically dead”. These are all examples of manufacturers
6 BARREL OPTIMIZATIONS 46
The optimization studied in this paper is an ”H” shaped cross section extrusion. From a
vibrations perspective, the Area Moment of Inertia is 2.9 times greater than the stock barrel in
the Tikka T3X system, and length and material modulus are exactly the same. This increase
in Area Moment of Inertia shifts the First Mode to 51.5 Hz, and the Second Mode to 319.8 Hz.
The following improvements to muzzle displacement, projectile exit angle, and vertical error at
Figure 44: Grid Style Extruded Barrel Blank Model, Area Moment of Inertia Optimization
6 BARREL OPTIMIZATIONS 47
Figure 45: Grid Style Extruded Barrel Blank, Modal Analysis, Mode 2
Inputing the moment of inertia improvement into the mathematical model, the response
comparison may be made as shown below. Note that Figure 46 shows total system response and
Figure 47 shows only the action time of the cartridge and system combination.
6 BARREL OPTIMIZATIONS 48
Figure 46: Moment of Inertia Optimization vs. Tikka T3X System Response
6 BARREL OPTIMIZATIONS 49
Figure 47: Moment of Inertia Optimization vs. Tikka T3X System Action Time Response
The improvements in muzzle displacement and angle at projectile exit represent an order of
magnitude improvement, which equates to significantly less vertical error at the target. The
geometry maximizes stiffness in bending and also lowers weight as compared to the nominal
rifle barrel. The trade off in this case is complexity in manufacturing. The barrel blank itself
may be extruded or forged on a mass production scale using standard manufacturing techniques.
However, follow on machining to fit the barrel to the receiver would likely be cumbersome.
Depending on the desired caliber of manufacture, and peak pressure of the round, the barrel
would have to gain cross sectional thickness to support the necessary loads at operating pressures,
which could be a detractor for weight eventually. The barrel blank rifling could be gun drilled
like a standard legacy rifle barrel, but not cold hammer forged, which is a much more common
6 BARREL OPTIMIZATIONS 50
commercial practice today. Still, it presents some interesting improvements strictly from an
A third improvement for rifle barrel vibration and precision is the concept of Modal Mass Tuning.
The Modal Mass matrix, or variable [µ] in the mathematical algorithm used to assess these
optimizations, is used in the calculations for system response. As stated above, the Modal Mass
T
[µ] = {ψ} [M ] {ψ} (40)
This results in mass applied to each degree of freedom proportional to the modeshape at
each mode. Thus, [µ] is not simply the total mass of the system divided by the displacement
degrees of freedom. It is a diagonal matrix, but the values of µii are not all equal. It may be
shown using the mathematical algorithm for this system’s response, that adding mass to the last
two degrees of freedom improve precision by reducing muzzle displacement and muzzle angular
error at projectile exit. However, if a tuning system were installed on the rifle that allowed
placement of masses at the longitudinal locations of the discretized masses of the barrel, it then
becomes possible to tune the response to something more favorable. For example, if the vector
ζ = [1, 1, 1, 5, 1, 5, 1, 5, 1, 5] were multiplied by the mass matrix M , the response of the system
entirely changes. This operation mathematically represents tuning masses placed at nodes 10, 8,
6, and 4 that sum to five times the discretized modal mass at those locations i.e. a series of masses
precisely mounted at nodal points along the vibrating beam. The impact in the mathematical
system is a significant improvement in both muzzle displacement and muzzle tip angle at the
The Modal Mass Tuning method of system optimization response may be compared to the
nominal response in the mathematical model easily. Using the parameters above, the following
plots were generated; the first for total system response through 0.1 seconds, and the second only
plotted through the action time of the cartridge and system being fired.
Figure 48: Modal Mass Tuning Scenario vs Nominal Tikka T3X Vibration Response
7 CONCLUSIONS 52
Figure 49: Grid Style Extruded Barrel Blank, Modal Analysis, Mode 2
As compared to the baseline performance in Table 2, modal mass tuning using this specific
mass arrangement, accounts for an 89% vertical error reduction from the mode 2 offset at 100
yards.
7 Conclusions
Regarding the topic of rifle barrel optimizations for muzzle tip displacement and projectile exit
angle, there are several optimizations that are possible. Within this paper, three distinct methods
of altering barrel response from the nominally offered product were reviewed. A rather simple
optimization, if the system and end user can tolerate reduction in muzzle velocity, is to reduce
the overall length of the barrel. By changing barrel length, the overall barrel stiffness is increased,
the resonant modes are shifted higher, and the muzzle tip displacement and angle at the moment
7 CONCLUSIONS 53
of projectile exit are favorable when compared to the nominal barrel length. For users who desire
a given barrel length for terminal performance, exploring a barrel cross sectional geometry that
provides an increased moment of inertia (2nd area moment), provides exceptional improvement to
the barrel dynamics. Specifically, the ”H” profile analyzed within this thesis offers very favorable
improvement in both muzzle tip displacement and projectile exit angle, which translates to
the reduction of error at the desired target range. In this case, users trade these favorable
characteristics with a potential increase in overall system weight (dependent on necessary barrel
wall thickness) and increased cost of manufacturing. Lastly, the concept of a tuning mass system
was explored, where masses were placed at nodes 10, 8, 6, and 4 to tune the barrel response.
This capability to tune a rifle barrel may be commercially adopted by the addition of sliding
masses on the barrel, with set screws. Alternately, a gear profile cut into a rib on the top of the
barrel could be used to mechanically increment masses longitudinally along the barrel to achieve
the same desired end effect. Analytical results of this technique demonstrated very favorable
response and muzzle angle results. Further, this type of system could be tuned to a particular
projectile mass and powder charge, increasing the potential for precision and reduction of error
at the target. The trade off is manufacturing cost, and overall complexity of such a system. It
is unlikely that such a system would be used by your average sportsman or military member.
Rather, it seems like a system for the bench rest shooter or precision rifle user.
In any case, these solutions explored are all very interesting, and my be desirable to the
commercial and military developers of precision firearm systems and high quality barrel suppliers.
REFERENCES 54
References
[1] Tikka Products, www.tikka.fi, 2022
[2] https://www.hunter-ed.com, Kalkomey, 2022
[3] http://ffden-2.phys.uaf.edu, M. Tilly, 2019
[4] https://www.luckygunner.com, Andrew, 2012
[5] ”Ballistics: Theory and Design of Guns and Ammunition”, 2nd Edition, D. Carlucci, S.
Jacobsen, 2013
[6] ”Structural Dynamics Theory and Application”, J. Tedesco, W. McDougal, C. A. Ross, 1999
[7] ”The Dynamical Behaviour of Structures”, G. B. Warburton, 1964
[8] ”Formulas for Natural Frequency and Modeshape”, R. Blevins, 2001
[9] https://hogsaddle.com/, Shadow Tech, 2022
[10] https://dewesoft.com/products/daq-systems/sirius, DEWESOFT, 2022
[11] https://training.dewesoft.com/online/course/, Sound Level Frequency Weighting Curves,
DEWESOFT, 2022
[12] ”Vibrations of Rifle Barrels”, A. Mallock, May 2, 1901
[13] ”Study on Barrel Vibration Characteristics of Typical Sniper Rifle”, Xu, Guan, Liu, Xu,
2018
[14] ”Concept of a Gun Barrel Based on the Layer Composite Reinforced with Continuous Fila-
ment”, Pyka et al., 2019
[15] ”Analysis and Control of Gun Barrel Vibrations”, F. Buyukcivelek, 2011
[16] ”A Review on the Gun Barrel Vibrations and Control for a Main Battle Tank”, Dursun,
Buyukcivelek, Utlu, 2017
[17] ”Dynamic Analysis of Gun Barrel Vibrations Due to Effect of an Unbalanced Projectile Con-
sidering 2-D Transverse Displacements of Barrel Tip Using a 3-D Element Technique”,
Koc, Esen, Cay, 2018
[18] ”Using Barrel Vibrations to Tune a Barrel”, G. Kolbe, 2021
[19] https://tacomhq.com/structured-barrels, 2021
[20] https://www.electronicshub.org/butterworth-filter, 2021
[21] ”Ammunition, Demystified. The (non) Bubba’s Guide to How Ammo Really Works”, J.
Siewert, 2022
[22] https://www.shootingsoftware.com/barrel.htm, 2004
[23] https://www.shootinguk.co.uk/guns/gun-maintenance/rifle-accuracy-87614, 2020
8 APPENDIX 55
8 Appendix
%this program calculates the stiffness matrix for a cantilever beam with 20
%degrees of freedom.
clear all
clc
close all
OD=.75; % in inches
ID=.22; % in inches
%____________________________________
%____________________________________
r1 = OD/2;
r2 = ID/2;
%system values
W = den*L*A; %lbs
m = W/386.4; % lbs*sec^2/in
%I = 3.2;
l = L/n; %in
dt=0.0001;
tmax=.1;
t=0:dt:tmax;
nsteps=length(t);
r=1+zeros(10,1);
for j=1:n
M(j) = m/n;
end
8 APPENDIX 56
M = diag(M);
% tune = [1 1 1 5 1 5 1 5 1 5];
% M = M*diag(tune);
%DOF1
k1=[24*E*I/l^3;-12*E*I/l^3;0;0;0;0;0;0;0;0;0;6*E*I/l^2;0;0;0;0;0;0;0;0];
%DOF2
k2=[-12*E*I/l^3;24*E*I/l^3;-12*E*I/l^3;0;0;0;0;0;0;0;-6*E*I/l^2;0;6*E*I/l^2;0;0;0;0;0;0;0];
%DOF3
k3=[0;-12*E*I/l^3;24*E*I/l^3;-12*E*I/l^3;0;0;0;0;0;0;0;-6*E*I/l^2;0;6*E*I/l^2;0;0;0;0;0;0];
%DOF4
k4=[0;0;-12*E*I/l^3;24*E*I/l^3;-12*E*I/l^3;0;0;0;0;0;0;0;-6*E*I/l^2;0;6*E*I/l^2;0;0;0;0;0];
%DOF5
k5=[0;0;0;-12*E*I/l^3;24*E*I/l^3;-12*E*I/l^3;0;0;0;0;0;0;0;-6*E*I/l^2;0;6*E*I/l^2;0;0;0;0];
%DOF6
k6=[0;0;0;0;-12*E*I/l^3;24*E*I/l^3;-12*E*I/l^3;0;0;0;0;0;0;0;-6*E*I/l^2;0;6*E*I/l^2;0;0;0];
%DOF7
k7=[0;0;0;0;0;-12*E*I/l^3;24*E*I/l^3;-12*E*I/l^3;0;0;0;0;0;0;0;-6*E*I/l^2;0;6*E*I/l^2;0;0];
%DOF8
k8=[0;0;0;0;0;0;-12*E*I/l^3;24*E*I/l^3;-12*E*I/l^3;0;0;0;0;0;0;0;-6*E*I/l^2;0;6*E*I/l^2;0];
%DOF9
k9=[0;0;0;0;0;0;0;-12*E*I/l^3;24*E*I/l^3;-12*E*I/l^3;0;0;0;0;0;0;0;-6*E*I/l^2;0;6*E*I/l^2];
%DOF10
k10=[0;0;0;0;0;0;0;0;-12*E*I/l^3;12*E*I/l^3;0;0;0;0;0;0;0;0;-6*E*I/l^2;-6*E*I/l^2];
%DOF11
k11=[0;-6*E*I/l^2;0;0;0;0;0;0;0;0;8*E*I/l;2*E*I/l;0;0;0;0;0;0;0;0];
%DOF12
k12=[6*E*I/l^2;0;-6*E*I/l^2;0;0;0;0;0;0;0;2*E*I/l;8*E*I/l;2*E*I/l;0;0;0;0;0;0;0];
%DOF13
k13=[0;6*E*I/l^2;0;-6*E*I/l^2;0;0;0;0;0;0;0;2*E*I/l;8*E*I/l;2*E*I/l;0;0;0;0;0;0];
%DOF14
k14=[0;0;6*E*I/l^2;0;-6*E*I/l^2;0;0;0;0;0;0;0;2*E*I/l;8*E*I/l;2*E*I/l;0;0;0;0;0];
%DOF15
k15=[0;0;0;6*E*I/l^2;0;-6*E*I/l^2;0;0;0;0;0;0;0;2*E*I/l;8*E*I/l;2*E*I/l;0;0;0;0];
%DOF16
k16=[0;0;0;0;6*E*I/l^2;0;-6*E*I/l^2;0;0;0;0;0;0;0;2*E*I/l;8*E*I/l;2*E*I/l;0;0;0];
8 APPENDIX 57
%DOF17
k17=[0;0;0;0;0;6*E*I/l^2;0;-6*E*I/l^2;0;0;0;0;0;0;0;2*E*I/l;8*E*I/l;2*E*I/l;0;0];
%DOF18
k18=[0;0;0;0;0;0;6*E*I/l^2;0;-6*E*I/l^2;0;0;0;0;0;0;0;2*E*I/l;8*E*I/l;2*E*I/l;0];
%DOF19
k19=[0;0;0;0;0;0;0;6*E*I/l^2;0;-6*E*I/l^2;0;0;0;0;0;0;0;2*E*I/l;8*E*I/l;2*E*I/l];
%DOF20
k20=[0;0;0;0;0;0;0;0;6*E*I/l^2;-6*E*I/l^2;0;0;0;0;0;0;0;0;2*E*I/l;4*E*I/l];
K = [k1 k2 k3 k4 k5 k6 k7 k8 k9 k10 k11 k12 k13 k14 k15 k16 k17 k18 k19 k20];
tf = issymmetric(K);
if tf == 1;
waitfor(msgbox(’K is symmetric’,’Success’));
else
return;
end
Ktt = K(1:10,1:10);
Kt0 = K(1:10,11:20);
K0t = K(11:20,1:10);
K00 = K(11:20,11:20);
[sai,lambda] = eig(Kt,M)
omega = sqrt(lambda);
freq = (omega/(2*pi))
%____________________________________________________________________
%____________________________________________________________________|
%calculate modal influence factor for first three modes ref: Blevins
%_____________________________________________________________________
%Dynamic Loading
counter = 1;
for j=1:nsteps
if t(j)<= at
phi_a2(j) = b*MSfac2;
F_a2(j) = (P*phi_a2(j))/MM(9,9);
phi_a3(j) = b*MSfac3;
F_a3(j) = (P*phi_a3(j))/MM(8,8);
else
phi_a1(j) = 0;
F_a1(j) = 0;
phi_a2(j) = 0;
F_a2(j) = 0;
phi_a3(j) = 0;
F_a3(j) = 0;
end
end
%response calculations
damp=(.05+zeros(10,1)); %damping at 5%
xi=damp;
omegad=sqrt(1-xi.^2).*omega’;
%h1=(1/(omegad(10,10)))*exp(-omega(10,10)*xi(1)*t).*sin(omegad(10,10)*t);
h1=(1/(omegad(10,10)*MM(10,10)))*exp(-omega(10,10)*xi(1)*t).*sin(omegad(10,10)*t);
z1=dt*conv(h1,sai(:,10)’*r*F_a1);
q1=sai(:,10)*z1;
%h2=(1/(omegad(9,9)))*exp(-omega(9,9)*xi(2)*t).*sin(omegad(9,9)*t);
h2=(1/(omegad(9,9)*MM(9,9)))*exp(-omega(9,9)*xi(2)*t).*sin(omegad(9,9)*t);
z2=dt*conv(h2,sai(:,9)’*r*F_a2);
8 APPENDIX 59
q2=sai(:,9)*z2;
%h3=(1/(omegad(8,8)))*exp(-omega(8,8)*xi(3)*t).*sin(omegad(8,8)*t);
h3=(1/(omegad(8,8)*MM(8,8)))*exp(-omega(8,8)*xi(3)*t).*sin(omegad(8,8)*t);
z3=dt*conv(h3,sai(:,8)’*r*F_a3);
q3=sai(:,8)*z3;
%total response
x(1,:)=q1(1,:)+q2(1,:)+q3(1,:);
x(2,:)=q1(2,:)+q2(2,:)+q3(2,:);
x(3,:)=q1(3,:)+q2(3,:)+q3(3,:);
%plotting------------------------------------------------------------
figure
plot(t,x(1,1:nsteps),’k’,’LineWidth’,2)
hold
plot(t,q1(1,1:nsteps),’b’)
plot(t,q2(1,1:nsteps),’r’)
plot(t,q3(1,1:nsteps),’g’)
figure
plot(t,x(2,1:nsteps),’k’,’LineWidth’,2)
hold
plot(t,q1(2,1:nsteps),’b’)
plot(t,q2(2,1:nsteps),’r’)
plot(t,q3(2,1:nsteps),’g’)
% figure
% plot(t,x(3,1:nsteps),’k’,’LineWidth’,2)
% hold
% plot(t,q1(3,1:nsteps),’b’)
% plot(t,q2(3,1:nsteps),’r’)
% plot(t,q3(3,1:nsteps),’g’)
%mode 1
m1node9 = sai(9,10);
m1node10 = sai(10,10);
pkdisp1 = max(x(1,1:counter))
m1slope = ((m1node10-m1node9)*pkdisp1)/(L/n);
8 APPENDIX 60
%mode 2
m2node9 = sai(9,9);
m2node10 = sai(10,9);
pkdisp2 = max(x(2,1:counter))
m2slope = ((m2node10-m2node9)*pkdisp2)/(L/n);
%mode 3
m3node9 = sai(9,8);
m3node10 = sai(10,8);
pkdisp3 = max(x(3,1:counter))
m3slope =((m3node10-m3node9)*pkdisp3)/(L/n);
angle_m1 = atand(m1slope)
angle_m2 = atand(m2slope)
angle_m3 = atand(m3slope)
sai =
0.6053 0.9645 -1.0000 1.0000 -0.9153 0.7315 0.4866 0.2805 0.1092 0.0163
-0.7825 -1.0000 0.6101 -0.0397 -0.5761 0.9835 1.0000 0.7625 0.3590 0.0624
0.9282 0.7768 0.1347 -0.8637 0.8826 -0.1470 0.6792 1.0000 0.6379 0.1336
-1.0000 -0.3053 -0.7928 0.7293 0.3804 -1.0000 -0.2687 0.7782 0.8492 0.2257
0.9948 -0.2614 0.8919 0.2962 -0.9603 -0.2581 -0.9234 0.1786 0.9215 0.3343
-0.9128 0.7462 -0.3632 -0.9603 -0.1977 0.8968 -0.6374 -0.4882 0.8169 0.4555
0.7604 -0.9971 -0.4210 0.4547 1.0000 0.6385 0.2855 -0.8579 0.5337 0.5856
-0.5496 0.9365 0.9110 0.6102 0.0179 -0.6081 0.8826 -0.7068 0.1021 0.7213
0.2916 -0.5709 -0.7397 -0.8958 -0.9446 -0.8059 0.4781 -0.0584 -0.4273 0.8601
8 APPENDIX 61
-0.0758 0.1586 0.2321 0.3394 0.4721 0.6077 -0.7163 0.8638 -1.0000 1.0000
lambda =
1.0e+09 *
2.1880 0 0 0 0 0 0 0 0 0
0 1.7351 0 0 0 0 0 0 0 0
0 0 1.1866 0 0 0 0 0 0 0
0 0 0 0.7097 0 0 0 0 0 0
0 0 0 0 0.3722 0 0 0 0 0
0 0 0 0 0 0.1675 0 0 0 0
0 0 0 0 0 0 0.0611 0 0 0
0 0 0 0 0 0 0 0.0158 0 0
0 0 0 0 0 0 0 0 0.0020 0
0 0 0 0 0 0 0 0 0 0.0001
freq =
1.0e+03 *
7.4446 0 0 0 0 0 0 0 0 0
0 6.6296 0 0 0 0 0 0 0 0
0 0 5.4824 0 0 0 0 0 0 0
0 0 0 4.2398 0 0 0 0 0 0
0 0 0 0 3.0704 0 0 0 0 0
0 0 0 0 0 2.0601 0 0 0 0
0 0 0 0 0 0 1.2437 0 0 0
8 APPENDIX 62
0 0 0 0 0 0 0 0.6322 0 0
0 0 0 0 0 0 0 0 0.2248 0
0 0 0 0 0 0 0 0 0 0.0357
pkdisp1 =
0.0016
pkdisp2 =
0.0055
pkdisp3 =
0.0107
angle_m1 =
0.0053
angle_m2 =
-0.0768
8 APPENDIX 63
angle_m3 =
0.2408
peak_disp_muzzle =
angles =
v_error_100 =
>>
8 APPENDIX 64
8 APPENDIX 65